Model Training Report
Generated for model: models/resnet50_prune20pct_best_model.pth
Model Architecture
Text Summary
----------------------------------------------------------------
Layer (type) Output Shape Param #
================================================================
Conv2d-1 [-1, 64, 112, 112] 9,408
BatchNorm2d-2 [-1, 64, 112, 112] 128
MaxPool2d-3 [-1, 64, 56, 56] 0
Conv2d-4 [-1, 64, 56, 56] 4,096
BatchNorm2d-5 [-1, 64, 56, 56] 128
Conv2d-6 [-1, 64, 56, 56] 36,864
BatchNorm2d-7 [-1, 64, 56, 56] 128
Conv2d-8 [-1, 256, 56, 56] 16,384
BatchNorm2d-9 [-1, 256, 56, 56] 512
Conv2d-10 [-1, 256, 56, 56] 16,384
BatchNorm2d-11 [-1, 256, 56, 56] 512
Bottleneck-12 [-1, 256, 56, 56] 0
Conv2d-13 [-1, 64, 56, 56] 16,384
BatchNorm2d-14 [-1, 64, 56, 56] 128
Conv2d-15 [-1, 64, 56, 56] 36,864
BatchNorm2d-16 [-1, 64, 56, 56] 128
Conv2d-17 [-1, 256, 56, 56] 16,384
BatchNorm2d-18 [-1, 256, 56, 56] 512
Bottleneck-19 [-1, 256, 56, 56] 0
Conv2d-20 [-1, 64, 56, 56] 16,384
BatchNorm2d-21 [-1, 64, 56, 56] 128
Conv2d-22 [-1, 64, 56, 56] 36,864
BatchNorm2d-23 [-1, 64, 56, 56] 128
Conv2d-24 [-1, 256, 56, 56] 16,384
BatchNorm2d-25 [-1, 256, 56, 56] 512
Bottleneck-26 [-1, 256, 56, 56] 0
Conv2d-27 [-1, 128, 56, 56] 32,768
BatchNorm2d-28 [-1, 128, 56, 56] 256
Conv2d-29 [-1, 128, 28, 28] 147,456
BatchNorm2d-30 [-1, 128, 28, 28] 256
Conv2d-31 [-1, 512, 28, 28] 65,536
BatchNorm2d-32 [-1, 512, 28, 28] 1,024
Conv2d-33 [-1, 512, 28, 28] 131,072
BatchNorm2d-34 [-1, 512, 28, 28] 1,024
Bottleneck-35 [-1, 512, 28, 28] 0
Conv2d-36 [-1, 128, 28, 28] 65,536
BatchNorm2d-37 [-1, 128, 28, 28] 256
Conv2d-38 [-1, 128, 28, 28] 147,456
BatchNorm2d-39 [-1, 128, 28, 28] 256
Conv2d-40 [-1, 512, 28, 28] 65,536
BatchNorm2d-41 [-1, 512, 28, 28] 1,024
Bottleneck-42 [-1, 512, 28, 28] 0
Conv2d-43 [-1, 128, 28, 28] 65,536
BatchNorm2d-44 [-1, 128, 28, 28] 256
Conv2d-45 [-1, 128, 28, 28] 147,456
BatchNorm2d-46 [-1, 128, 28, 28] 256
Conv2d-47 [-1, 512, 28, 28] 65,536
BatchNorm2d-48 [-1, 512, 28, 28] 1,024
Bottleneck-49 [-1, 512, 28, 28] 0
Conv2d-50 [-1, 128, 28, 28] 65,536
BatchNorm2d-51 [-1, 128, 28, 28] 256
Conv2d-52 [-1, 128, 28, 28] 147,456
BatchNorm2d-53 [-1, 128, 28, 28] 256
Conv2d-54 [-1, 512, 28, 28] 65,536
BatchNorm2d-55 [-1, 512, 28, 28] 1,024
Bottleneck-56 [-1, 512, 28, 28] 0
Conv2d-57 [-1, 256, 28, 28] 131,072
BatchNorm2d-58 [-1, 256, 28, 28] 512
Conv2d-59 [-1, 256, 14, 14] 589,824
BatchNorm2d-60 [-1, 256, 14, 14] 512
Conv2d-61 [-1, 1024, 14, 14] 262,144
BatchNorm2d-62 [-1, 1024, 14, 14] 2,048
Conv2d-63 [-1, 1024, 14, 14] 524,288
BatchNorm2d-64 [-1, 1024, 14, 14] 2,048
Bottleneck-65 [-1, 1024, 14, 14] 0
Conv2d-66 [-1, 256, 14, 14] 262,144
BatchNorm2d-67 [-1, 256, 14, 14] 512
Conv2d-68 [-1, 256, 14, 14] 589,824
BatchNorm2d-69 [-1, 256, 14, 14] 512
Conv2d-70 [-1, 1024, 14, 14] 262,144
BatchNorm2d-71 [-1, 1024, 14, 14] 2,048
Bottleneck-72 [-1, 1024, 14, 14] 0
Conv2d-73 [-1, 256, 14, 14] 262,144
BatchNorm2d-74 [-1, 256, 14, 14] 512
Conv2d-75 [-1, 256, 14, 14] 589,824
BatchNorm2d-76 [-1, 256, 14, 14] 512
Conv2d-77 [-1, 1024, 14, 14] 262,144
BatchNorm2d-78 [-1, 1024, 14, 14] 2,048
Bottleneck-79 [-1, 1024, 14, 14] 0
Conv2d-80 [-1, 256, 14, 14] 262,144
BatchNorm2d-81 [-1, 256, 14, 14] 512
Conv2d-82 [-1, 256, 14, 14] 589,824
BatchNorm2d-83 [-1, 256, 14, 14] 512
Conv2d-84 [-1, 1024, 14, 14] 262,144
BatchNorm2d-85 [-1, 1024, 14, 14] 2,048
Bottleneck-86 [-1, 1024, 14, 14] 0
Conv2d-87 [-1, 256, 14, 14] 262,144
BatchNorm2d-88 [-1, 256, 14, 14] 512
Conv2d-89 [-1, 256, 14, 14] 589,824
BatchNorm2d-90 [-1, 256, 14, 14] 512
Conv2d-91 [-1, 1024, 14, 14] 262,144
BatchNorm2d-92 [-1, 1024, 14, 14] 2,048
Bottleneck-93 [-1, 1024, 14, 14] 0
Conv2d-94 [-1, 256, 14, 14] 262,144
BatchNorm2d-95 [-1, 256, 14, 14] 512
Conv2d-96 [-1, 256, 14, 14] 589,824
BatchNorm2d-97 [-1, 256, 14, 14] 512
Conv2d-98 [-1, 1024, 14, 14] 262,144
BatchNorm2d-99 [-1, 1024, 14, 14] 2,048
Bottleneck-100 [-1, 1024, 14, 14] 0
Conv2d-101 [-1, 512, 14, 14] 524,288
BatchNorm2d-102 [-1, 512, 14, 14] 1,024
Conv2d-103 [-1, 512, 7, 7] 2,359,296
BatchNorm2d-104 [-1, 512, 7, 7] 1,024
Conv2d-105 [-1, 2048, 7, 7] 1,048,576
BatchNorm2d-106 [-1, 2048, 7, 7] 4,096
Conv2d-107 [-1, 2048, 7, 7] 2,097,152
BatchNorm2d-108 [-1, 2048, 7, 7] 4,096
Bottleneck-109 [-1, 2048, 7, 7] 0
Conv2d-110 [-1, 512, 7, 7] 1,048,576
BatchNorm2d-111 [-1, 512, 7, 7] 1,024
Conv2d-112 [-1, 512, 7, 7] 2,359,296
BatchNorm2d-113 [-1, 512, 7, 7] 1,024
Conv2d-114 [-1, 2048, 7, 7] 1,048,576
BatchNorm2d-115 [-1, 2048, 7, 7] 4,096
Bottleneck-116 [-1, 2048, 7, 7] 0
Conv2d-117 [-1, 512, 7, 7] 1,048,576
BatchNorm2d-118 [-1, 512, 7, 7] 1,024
Conv2d-119 [-1, 512, 7, 7] 2,359,296
BatchNorm2d-120 [-1, 512, 7, 7] 1,024
Conv2d-121 [-1, 2048, 7, 7] 1,048,576
BatchNorm2d-122 [-1, 2048, 7, 7] 4,096
Bottleneck-123 [-1, 2048, 7, 7] 0
AdaptiveAvgPool2d-124 [-1, 2048, 1, 1] 0
Linear-125 [-1, 1] 2,049
================================================================
Total params: 23,510,081
Trainable params: 23,510,081
Non-trainable params: 0
----------------------------------------------------------------
Input size (MB): 0.57
Forward/backward pass size (MB): 213.24
Params size (MB): 89.68
Estimated Total Size (MB): 303.50
----------------------------------------------------------------
Sparsity Analysis
======================================================== --- Custom Model Pruning Summary (Sparsity Analysis) --- ======================================================== Layer Name | Total Params | Non-Zero Params | Sparsity (%) ------------------------------------------------------------------------------------------------ conv1 | 9,408 | 8,706 | 7.46% layer1.0.conv1 | 4,096 | 3,869 | 5.54% layer1.0.conv2 | 36,864 | 31,262 | 15.20% layer1.0.conv3 | 16,384 | 15,491 | 5.45% layer1.0.shortcut.0 | 16,384 | 15,496 | 5.42% layer1.1.conv1 | 16,384 | 14,637 | 10.66% layer1.1.conv2 | 36,864 | 31,446 | 14.70% layer1.1.conv3 | 16,384 | 15,515 | 5.30% layer1.2.conv1 | 16,384 | 14,624 | 10.74% layer1.2.conv2 | 36,864 | 31,304 | 15.08% layer1.2.conv3 | 16,384 | 15,484 | 5.49% layer2.0.conv1 | 32,768 | 29,357 | 10.41% layer2.0.conv2 | 147,456 | 119,158 | 19.19% layer2.0.conv3 | 65,536 | 60,536 | 7.63% layer2.0.shortcut.0 | 131,072 | 116,917 | 10.80% layer2.1.conv1 | 65,536 | 56,108 | 14.39% layer2.1.conv2 | 147,456 | 119,213 | 19.15% layer2.1.conv3 | 65,536 | 60,541 | 7.62% layer2.2.conv1 | 65,536 | 56,197 | 14.25% layer2.2.conv2 | 147,456 | 119,319 | 19.08% layer2.2.conv3 | 65,536 | 60,627 | 7.49% layer2.3.conv1 | 65,536 | 56,010 | 14.54% layer2.3.conv2 | 147,456 | 119,499 | 18.96% layer2.3.conv3 | 65,536 | 60,650 | 7.46% layer3.0.conv1 | 131,072 | 112,401 | 14.24% layer3.0.conv2 | 589,824 | 459,206 | 22.15% layer3.0.conv3 | 262,144 | 234,201 | 10.66% layer3.0.shortcut.0 | 524,288 | 447,057 | 14.73% layer3.1.conv1 | 262,144 | 214,255 | 18.27% layer3.1.conv2 | 589,824 | 459,692 | 22.06% layer3.1.conv3 | 262,144 | 234,345 | 10.60% layer3.2.conv1 | 262,144 | 213,651 | 18.50% layer3.2.conv2 | 589,824 | 458,166 | 22.32% layer3.2.conv3 | 262,144 | 234,589 | 10.51% layer3.3.conv1 | 262,144 | 213,339 | 18.62% layer3.3.conv2 | 589,824 | 457,707 | 22.40% layer3.3.conv3 | 262,144 | 234,370 | 10.59% layer3.4.conv1 | 262,144 | 213,488 | 18.56% layer3.4.conv2 | 589,824 | 456,744 | 22.56% layer3.4.conv3 | 262,144 | 234,751 | 10.45% layer3.5.conv1 | 262,144 | 213,085 | 18.71% layer3.5.conv2 | 589,824 | 455,580 | 22.76% layer3.5.conv3 | 262,144 | 234,267 | 10.63% layer4.0.conv1 | 524,288 | 425,539 | 18.83% layer4.0.conv2 | 2,359,296 | 1,769,880 | 24.98% layer4.0.conv3 | 1,048,576 | 896,161 | 14.54% layer4.0.shortcut.0 | 2,097,152 | 1,698,605 | 19.00% layer4.1.conv1 | 1,048,576 | 816,573 | 22.13% layer4.1.conv2 | 2,359,296 | 1,772,108 | 24.89% layer4.1.conv3 | 1,048,576 | 895,453 | 14.60% layer4.2.conv1 | 1,048,576 | 816,822 | 22.10% layer4.2.conv2 | 2,359,296 | 1,766,830 | 25.11% layer4.2.conv3 | 1,048,576 | 893,129 | 14.82% linear | 2,048 | 1,608 | 21.48% ------------------------------------------------------------------------------------------------ TOTAL PRUNABLE | 23,456,960 | 18,765,568 | 20.00% ========================================================
Graphical Visualization
Visualisasi Sparsity Bobot
Heatmap from weight matrix for sample layer. White pixel represents pruned (zeroed) weights.
conv1
layer1.0.conv1
layer1.0.conv2
layer1.0.conv3
layer1.0.shortcut.0
layer1.1.conv1
layer1.1.conv2
layer1.1.conv3
layer1.2.conv1
layer1.2.conv2
layer1.2.conv3
layer2.0.conv1
layer2.0.conv2
layer2.0.conv3
layer2.0.shortcut.0
layer2.1.conv1
layer2.1.conv2
layer2.1.conv3
layer2.2.conv1
layer2.2.conv2
layer2.2.conv3
layer2.3.conv1
layer2.3.conv2
layer2.3.conv3
layer3.0.conv1
layer3.0.conv2
layer3.0.conv3
layer3.0.shortcut.0
layer3.1.conv1
layer3.1.conv2
layer3.1.conv3
layer3.2.conv1
layer3.2.conv2
layer3.2.conv3
layer3.3.conv1
layer3.3.conv2
layer3.3.conv3
layer3.4.conv1
layer3.4.conv2
layer3.4.conv3
layer3.5.conv1
layer3.5.conv2
layer3.5.conv3
layer4.0.conv1
layer4.0.conv2
layer4.0.conv3
layer4.0.shortcut.0
layer4.1.conv1
layer4.1.conv2
layer4.1.conv3
layer4.2.conv1
layer4.2.conv2
layer4.2.conv3
linear
Training & Validation Metrics
Loss Over Epochs
Accuracy Over Epochs
Precision Over Epochs
Recall Over Epochs
F1-Score Over Epochs
Final Test Set Evaluation
Area Under ROC Curve (AUROC): 0.9986
ROC Curve
Confusion Matrix
Grad-CAM Analysis
Visualizing model's attention focus on sample images.
Dataset - Real
Original
Grad-CAM
Original
Grad-CAM
Original
Grad-CAM
Original
Grad-CAM
Original
Grad-CAM
Dataset - Fake
Original
Grad-CAM
Original
Grad-CAM
Original
Grad-CAM
Original
Grad-CAM
Original
Grad-CAM
SynthBuster - Fake
Original
Grad-CAM
Original
Grad-CAM
Original
Grad-CAM
Original
Grad-CAM
Original
Grad-CAM
Extracted Frames - Real
Original
Grad-CAM
Original
Grad-CAM
Original
Grad-CAM
Original
Grad-CAM
Original
Grad-CAM
Extracted Frames - Fake
Original
Grad-CAM
Original
Grad-CAM
Original
Grad-CAM
Original
Grad-CAM
Original
Grad-CAM
